Human DevelopmentResearch Paper

2010/30Dealing with employment risk:

Policy options for emerging marketsSimon Commander

United Nations Development ProgrammeHuman Development ReportsResearch Paper

November 2010

Human DevelopmentResearch Paper

2010/30Dealing with employment risk:

Policy options for emerging marketsSimon Commander

United Nations Development Programme Human Development Reports

Research Paper 2010/30 November 2010

Dealing with employment risk: Policy options for emerging markets1

Simon Commander

Simon Commander is Managing Partner at the EBRD, IE Business School and Altura Advisers. E-mail:

scommander@alturaadvisers.com

Comments should be addressed by email to the author(s).

1 Thanks to Jeni Klugman for discussion and comments and to Ginette Azcona for help with data.

Abstract The paper looks at the experience of advanced economies in dealing with employment volatility. It examines in detail the impact of labour market institutions on equilibrium unemployment and the p ossible le ssons f or e merging ma rket e conomies tr ying to d esign p olicy f or d ealing w ith unemployment and a wider, growing demand for social protection from their citizens. Part of the paper concentrates on t he t ransition e conomies whose i nstitutional c ontext m ay b e r elevant t o other emerging markets. Some leading principles in policy design are elaborated that take into account some o f t he co mmon features o f em erging m arkets, notably a p rotected public s ector, large informal sectors and weak institutions. Keywords: employment, unemployment, labour market, emerging markets, social protection JEL Classifications: H53, J21, J65, J68 The H uman D evelopment R esearch P aper ( HDRP) S eries i s a m edium f or s haring recent research c ommissioned t o i nform t he g lobal H uman D evelopment R eport, w hich i s publ ished annually, and further research in the field of human development. The HDRP Series is a quick-disseminating, informal publication whose titles could subsequently be revised for publication as articles in professional journals or chapters in books. The authors include leading academics and practitioners from around the world, as well as UNDP researchers. The findings, interpretations and conclusions a re s trictly t hose of t he authors and do not necessarily represent t he vi ews of UNDP or U nited Nations M ember S tates. M oreover, t he da ta m ay no t b e consistent w ith th at presented in Human Development Reports.

1

1. Introduction

Most individuals’ welfare is determined by their place in the labour market and by the

efficiency with which they are individually matched to employment, as also by the

aggregate efficiency of that market. Fluctuations in income and employment that

accompany the business cycle remain substantial with unemployment also experiencing

considerable volatility. The recession that started in 2007 has, for example, seen sharp

movements in all these aggregates. At the extreme, the Spanish unemployment rate has

doubled to over 20% in less than 24 months! Elsewhere within the OECD, less dramatic

but still significant, changes to both employment and unemployment rates have been

induced with implications for earnings and income.

While less transparent, many of the same features are at work in developing

and/or emerging markets. Although fluctuations in open unemployment are generally

harder to observe – and sometimes a less meaningful indicator – the extent of

underemployment involving variation in hours worked and therefore in income, tends to

be substantial1. Certainly, the current downturn has exerted a large impact on

employment in a range of emerging markets, although the elasticity of employment to

output changes has varied very widely across countries2

1 Interestingly, the US under-employment rate, which includes those whose hours have been cut and those working part time for lack of full time positions, was 16.8% in February 2010 at a time when the unemployment rate was 9.7%. Recourse to adjustment to hours worked has been a feature of the current recession in a range of OECD countries, most notably, in Germany.

. A widely cited instance has

been China where job losses in the coastal regions in 2008-2009, linked to falling

demand for exports, have resulted in significant numbers of workers returning to the

2 For example, between 2008-2009 output fell by over 5% but unemployment increased by less than 1.5% in Mexico, whereas in Turkey unemployment increased faster than the fall in output over the same period. See ILO (2010)

2

countryside. Yet, even absent the current recession, the question of employment

volatility and, in particular, the issue of what to do about unemployment, has become an

increasingly prominent policy challenge. Indeed how to address employment volatility in

middle income emerging markets, where both income levels and institutions are

potentially capable of supporting an organised response, is the subject of this paper.

In the advanced market economies, employment volatility and its associated

individual risk has been widely addressed since the 1940s - particularly in Western

Europe - through a variety of welfare programmes, most notably, unemployment

insurance. Spending on social protection in most Western Europe countries now ranges

between 25-30% of GDP3. While the architecture and coverage of such programmes has

remained notably leaner in North America, even there the trend has been towards

providing more fall-backs in the case of job loss in terms of both generosity and

coverage. The current recession has also forced some further convergence. In the USA

discretionary social spending – including unemployment benefits - has accounted for

nearly 40% of additional fiscal or stimulus spending. In the G-20 grouping of countries

almost half have extended the duration of unemployment benefits in 2009-10 while over

a third have expanded the coverage of unemployment benefits4

3 This is when using Eurostat data where social protection comprises spending on sickness and health care, disability, old age, survivors, family/child benefits, unemployment, housing and social exclusion.

. By 2009 around 150

countries worldwide were actually operating some form of unemployment compensation

programme, including a large number of developing countries. Even so, while roughly

4 Note, however, that the share of the unemployed receiving benefits continues to vary very substantially across countries. In the USA and Canada only around 40% received benefits in 2008 compared to between 60-90% in the major Western European countries (excluding the UK where the share was closer to 50%), see ILO (2010).

3

half the world’s population in 2009 lives in countries with some form of unemployment

compensation, the average amount of compensation – defined as wage loss replacement -

remains small at around 10%5. Nevertheless, even if self-insurance and/or household

based fallbacks continue to be the dominant mechanism for coping with employment

shocks in developing countries6

This paper has several goals. It aims, first, to look at the experience of the

advanced market countries in dealing with employment volatility, paying particular

attention to the role of policies dealing with employment protection and unemployment

benefits. It considers the evidence regarding the impact of labour market institutions on

equilibrium unemployment and hence what sorts of lessons might be derived from

cumulative experience in these countries. In addition, it places this in the context of the

demand for policy changes induced by wider integration into global markets and the

associated risk that this has implied. The emphasis in the paper is mainly on to two sorts

of interventions; those relating to the provision of unemployment benefits – normally

financed on an insurance basis – and employment protection. In addition, special

attention is paid to a set of emerging markets – namely the so-called transition countries

of Central and Eastern Europe – that emerged from the planned economy with a more

developed institutional structure and commitment to employment stability than most

comparator countries at similar income levels. It is conjectured that these countries might

, there is a clear increase in the incidence of publicly

supported schemes for introducing or augmenting income/employment risk mitigation –

not least because of growing demand from citizens and other interest groups.

5 Vroman and Bruntseva (2009) 6 Not only is the replacement rate low, but coverage is often small. Data for Argentina, Brazil, China, South Africa and Turkey put coverage in the range of 7-12%, rising to around 25% in Russia.

4

offer insight into the ways in which labour market institutions affect the overall

performance of the labour market and hence offer some partial clues to their suitability

elsewhere in the emerging market world. The final part of the paper addresses the

question of what other emerging markets, such as China or Brazil, should do in terms of

designing labour market interventions and institutions that can help cope with

unemployment in ways that avoid many of the pitfalls experienced in the advanced

market economies and are also feasible from both a financing and institutional

perspective.

2. Mitigating employment risk

Explicit publicly supported programmes addressed to unemployment date back in Europe

to the end of the nineteenth century but in the main have emerged in the aftermath of the

two world wars. One argument has explained the timing not only in terms of avoiding a

repeat of the Great Depression of the 1930s, but also to a trade-off between providing

greater income protection to citizens while pursuing greater trade opening and exposure

to the world economy. For example, Rodrik (1998) has argued that larger governments

have resulted as the price for citizens accepting greater external risk. That risk – as

measured by volatility of the terms of trade and the product concentration of exports - has

in turn been argued to lead to greater volatility in domestic income and consumption. The

operative assumption is that larger government size will be associated with lower

volatility, for example by controlling a larger share of government resources or funding

counter-cyclical policies, such as unemployment benefits. Indeed, in the richer countries,

greater external risk has no statistically significant impact on government consumption

5

but rather spending on social security and welfare or transfers is correlated with external

risk. For countries with less robust institutions – including most emerging markets – the

prediction is that government size – as primarily measured by government consumption -

will tend to be larger the more open the economy. This is not because of openness per se

but because of the external risk that openness implies.

As a simple measure of external risk, Chart 1 indicates for a sample of emerging

markets that openness has clearly increased since the early 1980s and often very

significantly. A measure of income terms of trade risk that compares the 1980s with the

period since 2000 gives more ambiguous results but suggests that for some of these

emerging markets volatility – as measured by the standard deviation of the income terms

of trade - has also increased. In this context, it appears – in an echo of what occurred in

Europe post-1945 - that there has been a shift in the preferences of citizens towards

greater social insurance against external risk. While the exact causality is hard to

identify, it would appear that this shift traces to a combination of growth in incomes

(Wagner’s Law) – and with it the size of the middle class – but also the independent

effect of the growth in externally conditioned risk. This risk has, of course, been recently

accentuated by the impact of the 2007 recession and the slowdown in trade that resulted.

Public policy is increasingly been pushed to extend beyond compensatory transfers –

conditional or unconditional – for those facing persistent income inadequacy to

addressing the wider concerns of citizens for more protection against income and

employment variability and risk7

7 There is some debate over the evolution of the labour share, particularly in the advanced market economies. Rodriguez (2009) argues that the labour share has fallen in recent decades. While the factors behind this are complex, it is possible that trade

.

6

Table 1 reports some responses from the 2005/6 round of the World Values

Survey for a number of emerging markets. The table reports responses to two related

questions. The first asked about the respondent’s view of the desirability or otherwise of

democracy. Depending on whether the respondent considered democracy desirable or

not, they were then asked whether the provision of state aid for unemployment was a key

feature of democracy. Responses to the latter question were in ascending order between 1

and 10. For those considering democracy to be broadly desirable, the table shows that

the mean response ranged between 5 and 7.7. For a minority of respondents considering

democracy not to be desirable, responses had a broadly similar range, albeit with some

differences in the direction of change across countries. While these responses in the

World Values Survey are – it should be stressed - only very indicative, they do suggest

that citizens mostly consider state support for the unemployed to be an important feature

of democracy. Country level opinion surveys also tend to confirm the view that citizens

in an increasing group of countries view protection against employment risk as important.

Two strands of the argument so far have emerged. The first is that external risk

appears to have increased for a number of middle income emerging markets. The second

is that, for a variety of reasons, there appears to be an increased appetite among citizens

of some of these countries for publicly led abatement of this risk. There is also a further

dimension that is particularly relevant in East Asia, notably China. That is the current

need for rebalancing the economy and raising domestic consumption. The latter has been

kept in check by savings rates and consumption decisions that have been driven by low

insurance and absent public provision of services and fallbacks. In short, macroeconomic

opening has played a role. Data for the UK show no clear trend for last 35 years, but excluding top 1% shows a clear decline after 1989/90 (see Atkinson (2005)).

7

imperatives, as much as those emanating from long run structural change, make

consideration of how to deal with individual risk increasingly central to the policy

agenda. However, these pressures are also occurring in the continuing context of limits

(often quite severe) on institutional abilities and integrity. These limits materially affect

what can be done and how (and are dealt with in more detail in Section 7 below).

While there is some evidence that there has been an increase in risk and hence in

employment volatility in many emerging markets, it has also been argued that increased

external risk has been present in many OECD countries. There, this increased risk may

have interacted with existing labour market institutions in ways that have not necessarily

been beneficial. Thus, employment protection may have become more binding raising

costs to firms, while unemployment benefits may have also become more costly as more

people pass into unemployment. But the empirical evidence for the OECD countries

regarding the conjectured increase in risk is, however, not that convincing. Even so, the

interaction between institutions and shocks over time can have consequences for

employment that are long lasting and not necessarily beneficial. An example widely cited

is that the shocks of the 1970s and 1980s in Western Europe gave rise to a change in

institutions that has only been partially undone decades later with negative consequences

for employment8

Despite the common role that external factors may have played in raising the

demand for risk abatement, the model for achieving that purpose is far from obvious or

. But the possible longer term impact of policies designed for particular

episodes or shocks in the OECD suggests the need for very careful attention in design in

today’s emerging markets.

8 See Blanchard (2006), Ljungqvist and Sargent (1998)

8

indeed unique. In fact, of the many elements of the policy armoury in the advanced

market economies – a natural reference point - those relating to the labour market remain

ones where a persistent sense of failure in design is widely present. These perceived

failures may straddle many elements, including affordability – the sheer cost of providing

not only unemployment insurance - but also the associated costs of other interventions,

whether direct – such as training – or indirect, as through the externalities arising from

particular interventions, such as employment protection. And, of course, there are the

issues of the impact of publicly financed transfer programmes on individual incentives

where experience suggests that reduced incentives for job search, increased dependency

and even across-generation transmission of habits and tastes can result from flawed

design. In short, the demand for labour market-centred risk reduction policies is clearly

present, but how best to tailor design is still problematic. The next section looks at

experience with employment related risk reduction policies in the advanced market

economies – notably in Europe – and tries to extract some of the main lessons from these

rather heterogeneous experiences.

3. Institutions of the Labour Market

The core features of most OECD countries include elements of employment protection

and income protection, through unemployment insurance. Over the last 30 years there has

been widespread reappraisal of the design of policies addressed to the labour market and,

in particular, the unemployed in the OECD. The reasons for this reappraisal are clear. As

Table 2 shows, unemployment in much of the OECD increased substantially in the 1970s

and 1980s and mostly did not revert back to the low rates that existed in the 1960s. In

9

addition, large shocks – such as that to Finland or Japan in the 1990s and to the global

economy post 2007 – have led to further jumps in unemployment rates. However, there

has remained considerable heterogeneity in unemployment across countries and across

time.

Turning to employment, the picture is also again variable. In Europe alone, recent

employment rates have ranged from lows of around 55% in Italy and Spain to highs of

78% in Norway and Switzerland, respectively. Inactivity rates would similarly show

large variation. Insofar as there are patterns, those countries with high unemployment

rates are mostly those that also have high inactivity rates and low employment rates.

Disaggregating this information further by age and gender, the picture is also very varied.

Certainly, it appears that incentives for work of different types of individuals and families

– older males, females with children and so on – varies significantly across countries

being influenced by policies on, inter alia, retirement age, child care availability as well

as the structure of the tax system. Indeed, evidence from the OECD, suggests that

marginal tax rates for women when their husbands work has proven important in

explaining inactivity among women. In similar vein, benefit levels for workers at the

lower end of the earnings spectrum have affected choices over labour force participation.

3.1 Unemployment benefits

In response to upward spikes in unemployment, policy changes have been far from

uniform but have tended to have several common components. Aimed at reducing the

duration effects of unemployment benefits, one general direction of change has been to

limit the duration and level of benefits and/or make them more conditional, not least to

10

try and reduce dependency. Yet, no uniform configuration has emerged. For duration of

benefits, currently the USA, UK, Canada and Italy limit benefits receipt to between 6-8

months. By contrast, in Denmark workers can draw benefits for up to 48 months, while in

Spain, Portugal, France, Finland and Norway this can extend for up to 23-24 months.

Other material differences relate to the earnings base on which benefits are calculated –

some countries using net, some gross earnings, the level of payments for family members

and dependents, the extent of previous employment for eligibility, as well as whether

insurance for workers is compulsory or voluntary – the Scandinavian countries having

voluntary systems in contrast to all other countries in the OECD sample9

The UK is, more generally, an interesting instance of reform, as the changes that

have been introduced in the last 20 years have pushed the current system rather closer to

the North American model than the approach still applied through much of Western

Europe. By the early 2000s the UK unemployment rate had fallen to around 5% and UK

unemployment rates had been significantly lower than in the large continental countries,

such as France or Germany although generally higher than in Japan, Holland or the

Scandinavian countries. The share of long term unemployed (> 12 months) has fallen to

around 20% of total unemployed from a peak of over 40% in the mid-1990s. Parallel to

these developments, the number of claimants for unemployment benefits has fallen

sharply since the mid-1990s although this has been partly offset by a sharp rise in

incapacity benefits, part of which can probably be attributed to ‘gaming’ the benefits

system.

.

9 The OECD provides a very useful template of characteristics for 29 member countries, stylizing around a 40 year-old single worker without children and an uninterrupted employment record. See OECD, Benefits and Wages, Paris 2007

11

The principal elements of policy have been reductions in the generosity of

benefits, enhanced monitoring of claimants to stimulate job search and targeted

employment programmes for specific groups, such as the disabled, lone parents and the

long run unemployed. For low wage workers, earned income tax credits have been

applied, albeit with mixed results. In terms of generosity, benefits have been capped at

very low levels and have come more to resemble social assistance. However, for low

wage workers with particular family characteristics, replacement rates can still be

relatively high. For instance, OECD data show that net replacement rates for a married

couple with one earner and two children in both the initial and long run stage of

unemployment can approach 80% for those with relatively low ex ante earnings (defined

as 67% of the average wage).

In general, replacement rates have declined significantly over the past couple of

decades. Yet looking only at average replacement rates can be misleading. Take the

example of France; in 2007 the minimum benefit as a share of the average wage was

around 30% but the maximum benefit was 240%. By contrast, in the UK benefits have

been at 9% of the average wage. Indeed, when looking at net replacement rates across

countries, there is very large variation not only by country but also by family type. The

OECD’s disaggregation of net replacement rates for three family types with and without

children at different ex ante earnings levels brings out very clearly that for some family

types – mainly those with ex ante low earnings (67% of average wages) – replacement

rates can still be high. To give a sense of the variation: in the Netherlands, families

having average or above (150%) average wages with children can expect to get a

replacement rate of between 56-78%, those with low earnings would get replacement

12

rates of between 83-107%. This type of difference is common across many countries

ranging from Japan and the USA to many European countries. Some of this difference

also carries over into long run unemployment. In Japan low earnings families in long

term unemployment would receive replacement rates of between 53-108% as against 35-

76% for high ex ante earnings groups. These differences not only reflect distributional

considerations at work but also carry implications for work incentives.

Although replacement rates are critical, it is not just the level of benefits that

could be likely to influence unemployment but also their duration, coverage and the

strictness of enforcement. Calculations of benefits duration relating replacement rates for

longer term unemployed to initial unemployed also show large variation across countries

show that in most OECD countries, the replacement rate tends to fall sharply over time.

Information on both coverage and strictness is less easily available, but with respect to

the latter, it appears that closer monitoring of the unemployed and use of sanctions can be

important in determining whether benefits have an influence on unemployment.

Countries, like Denmark or the Netherlands, are often cited as instances where relatively

generous replacement rates sit alongside relatively strict enforcement, thereby mitigating

the potential for the first to affect unemployment10

While the picture that emerges is of tightening eligibility, lower generosity of

benefits and stricter enforcement, there is still enormous heterogeneity throughout the

OECD, and within Western Europe. For example, Sweden and Denmark look radically

different from the UK or France or Germany, while in the USA, state level variation

remains significant. As such, there is no single template that has been applied to benefits,

.

10 Nickell (2002)

13

even if a broad common philosophy of trying to limit disincentives for job search has

taken hold.

3.2 Employment protection

Alongside unemployment insurance and benefits, many OECD countries continue to

pursue policies of employment protection either through explicit legislation or de facto

policy. The balance between favouring more employment protection for incumbents, as

against providing fallbacks to job losers through unemployment insurance and social

assistance, varies quite substantially across countries. Boeri et al (2002) have argued that

employment protection and unemployment benefits act more as substitutes than

complements with a resulting inverse relationship between the two in Western Europe.

This has also been linked to the political economy and the power of particular lobbies or

vested interests in the labour market. Chart 2 relates the OECD measures of employment

protection to the net replacement rates offered in 2007. It can be seen that there is an ∩

shape with countries having higher employment protection tending to offer lower net

replacement rates. In addition, there are a small group of Anglo-Saxon countries that

have low employment protection and low replacement rates.

Employment protection measures – as well as contractual forms – have been

widely viewed as affecting the ease with which new entrants – younger workers – can get

into jobs. Such protection, by definition, tends to benefit largely incumbents. By affecting

the incentive for a firm to hire workers, employment protection can slow the speed at

which vacancies are filled. The evidence suggests that consequently employment

protection can adversely affect the duration of unemployment. However, differences in

14

employment protection appear to be unrelated to differences in unemployment rates

across countries.

Protection may also affect the ability of insiders to extract rents in wage

bargaining. Thus, while employment protection will tend to reduce volatility in the

labour market by reducing the firing rate, it will also tend to lower incentives for firms to

post new jobs. Indeed, in the OECD much of recent productivity growth appears to have

come from entry and displacement rather than from productivity growth in existing firms.

Yet, entry and displacement – the importance of rivalry – necessarily entails job losses

and worker displacement as well. Bassanini et al (2009) have also used firm level data

from OECD countries to examine the impact of employment protection on productivity

and investment. Both papers employ a ‘difference in differences’ where sectors are

distinguished in terms of their high and low reallocation sectors. Employment protection

adversely affects productivity more in high reallocation sectors. Further, using a panel of

EU firms, Leonardi and Messina (2009) find that employment protection legislation tends

to reduce investment and capital per worker, as well as labour productivity, quite

significantly in high relative to low reallocation sectors. Bartelsman et al (2010) have also

argued that differences in innovation may also be explained by differences in

employment protection insofar as higher protection reduces the incentive for firms to

adopt more risky technology by raising the cost of separating workers. These

considerations and the associated emergence of dual labour markets have led to

reappraisal of the role of employment protection, even if the evidence for employment

protection having an important and adverse impact on unemployment is far from

conclusive.

15

A closer look at the OECD employment protection scores suggests that there is

considerable heterogeneity across countries and time11

The introduction of differentiated contracts using temporary contracts has largely

been motivated by the difficulties in reforming the existing systems of employment

protection and the power of incumbents and their lobbies

. For example, countries such as

Spain, Greece and France have markedly higher employment protection than the USA,

Canada or the UK while the Netherlands, Sweden and Denmark have intermediate scores.

As regards trends, the evidence points to declining protection in most countries, although

in a number of places with high protection - notably France, Norway and Spain - the

decline has been relatively small. However, this has camouflaged some important

nuances. A notable instance has been the proliferation of temporary contracts and

associated dual labour markets. France and Spain offer two clear instances of where

attempts have been made to introduce greater flexibility in a segment of the labour

market. New hires with temporary contracts have faced significantly different

employment contracts than those workers on permanent contracts.

12

11 A recent update of these measures can be found in Venn (2009).

. French fixed term

employment rose to around 14% of total employees by 1998 and has stayed at around

that share thereafter. In Spain the growth in temporary contracts has been even more

rapid in the 1990s, resulting in temporary employment accounting for between 31-33% of

total employment in the period from 1998 to 2007. This process was important in driving

the reduction in the Spanish unemployment rate from over 22% in 1994 to 8% in 2007.

However, the subsequent recession has seen a very sharp upward shift in Spanish

unemployment towards 20% by 2010, a significant part of which can be accounted for by

12 Saint-Paul (2000)

16

the non-renewal or otherwise of temporary contract workers. Spanish experience

suggests that the introduction of temporary contracts led to increased volatility of

employment by increasing both hiring and firing rates13. The increase in the hiring rate

cut into unemployment and into long run unemployment in particular. However, lower

firing costs have also facilitated large separations since 2007. Other consequences of

fixed term contracts have been a lower incentive for employers to invest in on-the-job

training, a higher wage premium for more educated workers on permanent contracts, as

well as impeding spatial mobility due to uncertainty14

The impact of the current recession on a dual labour market suggests that how

temporary contracts are implemented can also matter. A comparison of France and Spain

since 2007 found that Spanish unemployment would have been raised by 2.5 percentage

points less if Spain had had less stringent French employment protection legislation

.

15

Use of temporary contracts has been most pronounced in Western Europe.

Despite the fact that the problem of employment protection for incumbents and the

impact on job creation is a pressing issue for many developing countries, explicit use of

temporary contracts has been very limited. What is familiar, however, is the widespread

inability to address head-on the extent of employment protection and wage benefits

.

More generally, there has been questioning of the whole strategy of temporary contracts

with emphasis placed on replacement by a single labour contract alongside severance pay

conditioned on tenure.

13 More generally, see also Boeri and Garibaldi (2007) 14 See Bentolila, Dolado and Jimenez (2008) 15 Bentolila, Cahuc, Dolado and Le Barbanchon (2009) who also note that unemployment benefits have been fairly similar in both France and Spain, as have wage bargaining institutions.

17

offered to workers in the covered or formal sector. Even where this is recognised to be a

barrier to employment growth, policy makers have mostly preferred to use other

instruments, such as wage subsidies, to try and boost employment. In Tunisia, a highly

regulated labour market and high labour supply has resulted in significant and growing

unemployment, but policy has mainly been the use of wage subsidies and other active

labour market policies, as well as trying to raise firm entry rates16

. Both set of policies

have been costly and have had very limited results. However, European experience with

tow-tier labour markets is unlikely to be an appropriate model for developing countries

that are trying to change the extent of employment protection. We return to some of these

issues in Section 5 below.

3.3 Trade Unions

Trade unions have been important features of the institutional landscape in Europe. Their

impact has primarily been through the structure of wage bargaining and the result wage

levels. In this process, union membership, coverage and the locus of bargaining have

been generally viewed as relevant.

In terms of membership, it is widely held that unions have declined – as has

labour militancy – over the last forty years. Yet, the simplest time series measure of

density – defined as the share of union members in total employees – actually shows a

decline between 1960/64 and the end of the 1990s in eight Western European countries

but an actual increase in eight other countries17

16 World Bank (2008)

. It also shows large variation in the extent

of changes across these countries. However, Visser’s (2006) revised series of density data

17 Nickell (2002)

18

for the period 1970-2003 show not only very large variation in union rates across

countries but also a clear secular decline in most countries over this period. By 2003

union density rates were lower than in 1970 in all but four European countries – namely,

the Scandinavian countries and Belgium. In some instances, the decline has been very

large – from nearly 22% in France to around 8%, from 45% in the UK to 30%, from 63%

to around 35% in Austria. Similar falls in membership have occurred in other OECD

economies, such as the USA, Canada, Australia and Japan, with particularly large falls in

Australia and the USA18

While density rates are indicative of the presence of unions, they are also

potentially misleading, as the size of membership may be a poor measure of the actual

effect of unionisation. Coverage – defined as the share of workers whose employment is

affected by collective agreements negotiated between employers and unions - can be very

different from density. Most striking is the case of France where the latter was around 8%

in 2003 while coverage was 95%. Indeed, coverage data for most of Western Europe

show rates of between 80-95%, except in Germany (63%) and the UK (35%).

Decentralised bargaining in the UK accounts for this low coverage rates, as it also does

for the USA, Canada and Japan. But elsewhere in Europe multi-employer bargaining and

policies tend still to extend contractual outcomes to non-unionised firms. The available

. Unionisation rates have fallen particularly strongly in the

transition economies, such as Hungary, Poland and the Czech Republic, driven mainly by

changes in the political system. For all these countries, changes since 2003 have been

incompletely monitored but it seems plausible to assume that unionisation has continued

to decline.

18 Unionisation rates vary a good deal across US states; for example, being around 13% in New Hampshire and Iowa as against around 6% in South Dakota and Louisiana

19

evidence also suggests very different incidences of unionisation across private and public

sectors – with the public sector having generally far higher density rates – as well as

differences across groups of workers, whether cut by age, sex or type of contract. In the

UK 2007 data show that while 60% of public sector workers remain unionised, this has

fallen to around 16% for private sector workers. Further, over 70% of workers in the

public sector have their earnings governed by collective agreements as against 20% in the

private sector. In short, much of the general decline in unionisation has, indeed, occurred

in the private sector. Yet, factors such as coverage and sectoral incidence help explain

why despite falling membership, unionisation has continued to influence unemployment

through wage bargaining (see Section 4 below). In addition, collective bargaining can

materially affect employment protection in ways that differ from legislation19

.

3.4 Labour taxation

Financing for labour market programmes varies widely both across the advanced market

economies as well as in emerging and transition countries20

19 Venn (2009) documents collective bargaining about employment protection. The paper notes that in a number of countries, including Denmark, collective agreement provisions are significantly more generous than those in the legislation

. With respect to

unemployment benefits, for instance, in the Netherlands insured persons’ contributions to

a general unemployment fund average 3.5% of covered earnings. Employers contribute

around 5.8% of covered payroll to both the unemployment fund and a redundancy

payment fund. These payments also cover sickness and maternity benefits. By contrast,

in the UK unemployment benefits are financed out of employer and employee

20 For detailed individual country descriptions, see the website for ‘Social Security Programs Throughout the World’ – www.ssa.gov/policy/docs/progdesc/ssptw

20

contributions that also cover sickness and maternity benefits, work injury benefits, as

well as a specific contribution to the National Health Service. There is no earmarking of

unemployment financing. Similar differences in financing arrangements exist in

transition and emerging markets. In Poland, coverage of employed workers for

unemployment benefits is financed by employers’ contributions of 2.45% of payroll, with

workers not explicitly contributing for unemployment. Deficits are covered by the

government budget. In the Czech Republic, both employers and employees contribute

explicitly to unemployment insurance, while in Russia unemployment benefits are

financed from federal and local government budgets rather than from payroll taxes. In

Colombia, employers contribute 8.3% of annual salary with no contribution from

employees. In Chile, individual severance accounts have been established with the

insured contributing a base of 0.6% of monthly earnings and employers contributing

between 2.4-3% of covered payroll depending on a worker’s contract.

There is considerable debate over the impact of labour taxes – including the

payroll taxes that commonly finance unemployment insurance - on unemployment. Taxes

that treat income equally whatever its source (earnings or benefits) should not affect the

cost of labour to firms and hence not affect unemployment. Theoretically, there are

grounds for assuming that taxes that drive a difference between take home and cost of

labour to the firm – the so-called tax wedge - may have a negative impact on

employment. That would be true if taxes raised labour costs and affected firms’ demand

for labour. It could also be possible that different taxes had different effects. In the latter

regard, evidence suggests that there is no significant differential impact of payroll,

income or consumption taxes – the standard components of labour taxation – on labour

21

costs and hence on employment. However, when turning to the impact of total labour

taxation on employment the evidence appears more ambiguous.

Cross-section studies using OECD data have indicated that an increase in the tax

wedge can raise real labour costs21 and adversely affect employment and raise

unemployment. A 10% increase in the tax wedge has been estimated to reduce the

employment rate by between 0.3 - 2.7%, while the effect on the unemployment rate has

been between 1 - 2%. Nickell et al (2003) use a fixed effect panel which similarly finds

an increase of just over 1% in unemployment. An interesting regional study for Sweden

looking at whether regional differences in payroll taxes translated into differences in

employment has found no significant effect, although it is possible that lower tax rates

may have had a positive impact through encouraging entry of new firms. A large

reduction in the payroll tax rate did, however, have an unambiguous impact on wages as

employers and workers appear to have treated the reduction as a windfall to be shared,

mainly to the benefit of employers22

21 For example, Daveri and Tabellini (2000), Nickell et al (2002), Scarpetta (1996), Nickell and Layard (1999), Nicoletti and Scarpetta (2001)

. There is also evidence from Europe that tax

incentives can materially affect individuals’ at the margin of being employed, more than

they may affect the amount of hours worked by those already in work. Further, major

effects of the tax wedge are only likely to occur for low wage workers where additional

contributions cannot be shifted to workers and hence lead to an increase in costs. To

address these concerns, there has been increasing experimentation with in-work tax

22 Bennmarker, Mellander and Ockert (2008)

22

credits, particularly in the UK where there is some evidence that they have helped induce

some types of workers to take up employment23

Studies from non-OECD settings also paint a mixed picture. Gruber (1997) used

Chilean data and found that a 1981 reduction of payroll taxes by 25% was almost fully

shifted on to wages with no significant employment effects. A more recent exercise using

Argentine data for the period from 1995 to 2001 found that changes were only partly

shifted to wages and that there was no significant effect on employment. By contrast,

using the case of a large increase in payroll taxes in Colombia after 1993, Kugler (2009)

found that 20% of the increase was passed on to workers in the form of lower wages – a

finding consistent with Heckman and Pages’ (2004) wider study for Latin America which

found a 33% shifting. In the Colombian instance, the impact of a 10% rise in payroll

taxes was to lower formal employment – particularly of production workers - by up to

5%. On balance, the empirical evidence suggests that labour taxation can affect

employment but not necessarily by large magnitudes.

.

Turning to the actual evolution of labour taxation, the data show that in Western

Europe there has been a substantial increase in the tax wedge since the 1960s. Between

1960 and 1964 the average tax wedge in Western Europe was around 38% with a range

of 19% (Spain) to 57% (Italy). By 1996-2000 it had risen to over 53% with a range of

33% (Ireland) to 77% (Sweden)24

23 The extent to which this is the case depends very much on design. A review of the UK’s Working Families Tax Credit indicated that employment of lone mothers did increase but that was not the case for low income couples where incentives actually deteriorated, see Brewer (2009)

. Subsequently, there have been attempts to reduce

labour taxation, so that in 2008 the average rate had fallen to around 42% for the same 15

24 CEP/LSE database

23

European countries25. In the USA and Canada the total tax rate in 2008 was slightly

below or the same as their 1960-64 rates at around 30-31% while in Japan the tax rate

had climbed from 25% to 29.5% by 200826

While much of the increase in the wedge can be attributed to increases in personal

income tax, social security contributions by employers and employees constitute the

largest component for many countries. In Belgium and Germany, for example, employer

and employee contributions to around 34% of labour costs or between 60-66% of the

total tax wedge, while in France contributions comprise nearly 40% of labour costs, as

against 10% for income tax. In Denmark, personal income tax is the largest component at

over 30% with contributions accounting for only 11%. In the so-called Anglo-Saxon

countries, contributions tend to be lower than the OECD average being in the range of

14%-18% for the UK, North America and Canada and falling to under 6% for Australia.

. However, the general picture again remains

one of considerable heterogeneity across countries. At present, a country like Ireland has

a tax wedge below 23% while Belgium’s is 56%.

4. Some consequences of labour market institutions

The growth in OECD unemployment – as well as its persistence – in and after the 1970s

has spawned considerable interest in understanding how the equilibrium rate of

unemployment has been affected by institutions, shocks and other factors. The operative

assumptions have been that the equilibrium level of unemployment will be affected by

any variable that influences the ease with which unemployed individuals can be matched

25 The countries are Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and UK. 26 The 2008 numbers come from OECD (2008).

24

to available job vacancies, as well as by any variable that affects the way in which wage

bargaining occurs. Variation in unemployment across countries after 1980 appears to

be well explained by differences in labour market institutions27. A series of papers have

also tried to explain the time series variation in employment28

Blanchard and Wolfers (2000) have also used five year averages with baseline

variables and interactions to argue that it has not so much been institutions by themselves

but the interaction of institutions and shocks that has been behind persistence of high

unemployment in Western Europe. Thus, shocks – such as in the 1970s – may have lead

to changes in institutions, such as increases in the generosity of benefits. These in turn

may have contributed to the persistence of high unemployment leading to further

institutional adaptation, albeit not necessarily a return to previous institutional

arrangements. These adaptations, it has been argued, can develop their own dynamic;

. Nickell et al (2005) used

annual data for the OECD and include variables that could explain deviations of

unemployment from its equilibrium level. Such variables include aggregate demand,

productivity and wage shocks. In addition, they include time series measures of

institutional variables, including, inter alia, for employment protection, employment

taxes, replacement rates and benefits duration. They argue that over half of the upward

shift in equilibrium unemployment over the period 1960-1995 can be explained by

changes in institutions. In particular, they found that (in order of importance) the benefits

system; labour taxes, unions and changes in laws for employment protection have been

contributory factors.

27 See Layard, Nickell and Jackman (1991). See also the papers in Snower and de la Dehesa (1997) 28 Aside from those directly cited in the text, see also Belot and van Ours (2001), Baker et al (2005), Bertola et al (2005)

25

examples being the recourse to dual labour markets and contracts in parts of Continental

Europe29

In short, despite the evidence that institutions play a role in explaining differences

in unemployment and changes in unemployment, it is difficult, if not impossible, to

isolate a particular configuration of institutions that tends to be consistent with better

economic performance, particularly with respect to equilibrium unemployment.

Consequently, Freeman (1998) has argued that there appears to be no unique institutional

path to be followed. This agnostic conclusion contrasts with the apparently strong

presumptions about the sort of labour market institutions and extent of labour market

flexibility that would be optimal in, for example, the World Bank’s ‘Doing Business’

indicators. Indeed, this form of labour market liberalism has been widely propagated as

the model for emerging markets to follow. Certainly, this contrasts significantly with the

existing situation in most emerging markets where employment protection and dual

labour markets are widely present. What is very different, however, from the OECD

setting is the absence – except in the former socialist or transition countries - of organised

fall-backs for workers losing their jobs. The next section turns to describing the situation

currently existing across a swathe of emerging markets.

.

5. Emerging markets and their labour markets

Most emerging markets have relied on policies of employment protection and severance

as their main policy instruments. Indeed, employment protection has been a very

pronounced in many developing feature with relatively small formal sector workforces

29 As argued in Blanchard (2006), Blanchard and Landier (2002)

26

being sheltered by job protection and significant constraints on both hiring and firing. As

such, workers in such formal sector firms – often in turn in the public sector - tend to be

privileged not only in terms of earnings and working conditions, but also in terms of job

security and tenure. The pervasive problem of youth unemployment – a major feature, for

example, of the Arab countries – can be traced not only to supply side factors – high

labour force growth rates – but also to the limit on job creation resulting from

employment protection of relatively small numbers of incumbents30

Despite some serious caveats about the quality of measurement and coverage

.

31,

the Doing Business employment indicators provide some evidence concerning the extent

of employment protection. For example, in 2007 firing costs expressed in weeks of salary

were, 186 in Egypt, 108 in Indonesia, 91 in China, 87 in Vietnam, 56 in India, 37 in

Brazil and 24 in South Africa. To put this in context, most Western European countries,

as well as USA, Canada, Japan, Australia and New Zealand had severance costs under 20

weeks, albeit with some major exceptions, such as Portugal (99), Germany and Greece

(both 69). Using the same source’s rigidity of employment index32

The OECD overall employment protection indicator for 2008 allows comparison

of additional emerging markets with the core OECD group (see Table 3). The Western

European (excluding the Accession countries) average is around 2.05 while Canada, UK

, the picture is more

mixed with many advanced economies also having quite high levels of employment

rigidity.

30 For an overview of developments in the Middle East and North Africa, see Salehi-Isfahani (2010) 31 Commander and Tinn (2009) 32 This being an average of three sub-indices – difficulty of hiring and firing and rigidity of hours indices.

27

and USA range between 0.2-0.75. Japan and Australasia have scores intermediate

between Europe and the Anglo-Saxon group. The group of emerging markets have

significantly higher employment protection than the European average, with some

exceptions such as South Africa and most of the transition economies. What is evident is

that employment protection continues to be an important element of policy in many of the

leading emerging markets. Where time series evidence is available, it suggests that

reductions in employment protection have been more elusive in recent years in emerging

markets than in the advanced market economies. This persistence has co-existed

alongside dual labour markets in many of these economies so that relatively small shares

of the labour force actually benefiting from the relatively high levels of employment

protection. We return to this issue in Section 5 below.

What are the consequences of employment protection policies? There is relatively

little evidence of the impact on measures of performance. Besley and Burgess (2004)

have used state level Indian data and find that a wider body of labour regulations

adversely affects firm level performance. Botero et al (2004) use a larger cross country

dataset to estimate the impact of an index of labour regulations including restrictions on

firings and find that regulation raises unemployment, lowers labour force participation

and raises the size of the informal or unofficial economy. There is also, of course, a large

body of country level evidence from a wide swathe of developing countries over a

protracted period of time suggesting that the existence of a covered or protected sector

has been associated with larger informal sectors, job queuing and wait unemployment33

33 This sort of setting is, of course, central to Harris-Todaro models as well as later models of segmented labour markets. Note, however, that enforcement of labour regulation may make de jure and de facto regulation diverge substantially. Gimpelson

.

28

A relatively small number of emerging markets have already established the

beginnings of a system of unemployment benefits. For example, in Turkey and Korea

compulsory unemployment insurance has been put in place. Workers have specified

contribution and employment requirements for eligibility with benefits payable for

between 7-10 months. The replacement rate in both countries is 50% of the earnings

base; gross earnings in the case of Korea, net in the case of Turkey. Social security

contributions amount to around 16% of labour costs in Korea and 29% in Turkey. In

China, unemployment benefits are available for a relatively small share of the urban

labour force. Benefits are set by local governments at levels lower than the local

minimum wage and are payable for periods that depend on pre-existing employment

experience. The installation of unemployment benefits has been particularly widespread

in the former socialist or transition countries (which are addressed in more detail below).

Across regions and countries, the beneficiaries of unemployment benefits remain

a relatively small share of the total unemployed34. One estimate suggests that in Latin

America around 20% of the unemployed receive some form of unemployment

compensation with this proportion falling to 2/3% in North Africa, the Middle East and

Sub-Saharan Africa. Not only is coverage very limited but – where available – low

replacement rates dominate35

and Kapelushnikov (2009) provide evidence from Russia of large variation in enforcement across regions or oblasts.

. In some countries, such as Brazil and China – as in the

USA – there is in addition significant regional variation as the provinces play a major

part in determining statutory provisions for benefits.

34 Of course, there remain issues with the measurement of the unemployed in the first place. 35 See Vroman and Brusentsev (2009)

29

6. Transition economies

The transition countries are particularly interesting as for historical reasons they have had

particularly developed welfare programmes with explicit unemployment insurance and

fallbacks for job losers. Although most of these countries – particularly in Central Europe

– have income levels at the higher end of the emerging market distribution36

To date, there has been a relatively small literature that has looked at the design and

impact of labour market institutions in these countries. Boeri and Terrell (2002) argued

that relatively generous non-employment benefits in Central and Eastern Europe

established a wage floor that forced the least productive firms to shed workers and hence

implement restructuring. By contrast, in the CIS benefits were less generous, wages more

flexible and this – through much of the 1990s – was associated with labour hoarding and

downward wage flexibility but also less restructuring. Consequently, unemployment

emerged earlier and at higher incidences in Central and Eastern Europe than in countries

further east.

, they

nevertheless provide interesting experimental terrain for considering what the impact of

labor market institutions might be in other emerging markets.

Empirical evidence on the direct effect of institutions on unemployment is mixed.

Boeri and Burda (1996) found a small but significant effect of expenditure on active

labour market programmes, job creation, and programme intake on outflows from

unemployment into employment in the Czech Republic. Ham, Svejnar and Terrell (1998,

1999) also found that the institutional system had only a moderate effect on the duration

36 For example, by 2006 in PPP terms Hungary’s per capita income was roughly double that of Brazil and 3.5 times that of China. However, a country like Ukraine had an income level below that of Brazil and 30% higher than that of China while Russia was roughly comparable to Mexico.

30

of unemployment for both women and men in the Czech and Slovak Republics. In

addition they found some evidence of country variation - there was a greater

responsiveness to the unemployment compensation system in the Czech Republic. For

Poland, Puhani (2000) did not find any significant effect on unemployment duration

using information on changes in the duration of unemployment benefits in Poland.

Similarly, Micklewright and Nagy (1996) found little effect of changes to the benefit

system on duration in Hungary. However, they also found that over time an increasing

proportion of the unemployed searching for jobs actually received no financial support

from the state while an increasing proportion of those in receipt of benefits did not

search.

In terms of trends, the data suggest several evolutions in institutions over time. First,

the generosity of unemployment benefits has tended to fall, often quite significantly.

Second, for employment protection, it appears that there has been a decline in most

transition countries. However, there is a considerable country level heterogeneity. As

Table 3 shows, many transition countries have employment protection scores comparable

to Western Europe, and lower than Brazil, China or India37

37 See Venn (2009)

. However, while the

employment protection indicator for Russia is relatively low – equivalent to that of

Finland or the Netherlands - de jure and de facto employment protection varies

substantially in Russia. Data on layoffs, as well as the widespread use of wage arrears

and involuntary hours adjustment, suggest that labour hoarding is pervasive (not only in

Russia but also in Ukraine and other CIS countries), even if de jure employment

protection has decreased. Third, regarding funding, the Central European transition

31

countries tend to have a total tax wedge of around 40% of which by far the greater part is

composed of social security contributions. The total labour tax rate is thus quite similar to

Western Europe. In the CIS it seems that the total tax wedge has fallen.

Unemployment in the transition countries has been driven since 1990/92 by a series

of powerful restructuring and reallocation processes. Table 4 gives the unemployment

rate for the major countries at four year intervals. It can be seen that in most countries,

unemployment rose sharply near the start of transition and in a number of cases persisted

at high levels – such as in Poland - before declining sharply after 2004. In the CIS – as

given by Russia and Ukraine – unemployment rose less sharply. In short, these

economies have all experienced large increases in unemployment but with some notable

variation across countries and regions.

In exploring whether employment and unemployment rates can be related to

institutional variables, Lehmann and Muravyev (2010) run a set of cross country

regressions relating in the spirit of Nickell et al., (2005). They find that employment

protection in particular exerts a significant negative effect on the employment rate and a

positive effect on unemployment that is, however, insignificant, except when having

youth unemployment on the left hand side. Most of the other explanatory variables lack

significance and/or are perversely signed. The finding that employment protection acts

adversely against employment is consistent with findings for both the OECD and from

other emerging markets.

32

6.1 Unemployment and unemployment benefits in transition economies

Commander and Heitmueller (2008) look more systematically at whether unemployment

benefits have any impact on employment and unemployment. They show that for both

initial and long run unemployment, replacement rates in Central Europe and Russia

tended to fall between the mid 1990s and 2005. At that time, initial replacement rates

then ranged between 0.4-0.6. More recent – 2008 - indicators reported in Table 5 suggest

that the replacement rate for two family types averaged around 55% and were not that

different from the comparator advanced market economies also included in the table for

both initial and long term unemployment.

However, there are some important caveats. Not all the unemployed have actually

been eligible for unemployment benefits. Indeed, the share of those eligible has often

declined to low levels. For example, in Hungary and Poland eligibility rates of around 0.6

in the early/mid-1990s had fallen to around 0.2 by 1996/98. In the Czech Republic

eligibility has remained at around 0.2/0.3 from the mid-1990s to now38

38 Vroman and Brusentsev (2009) report recipiency rates for 2004/2005 for 10 Central and Eastern European countries. With the exception of Romania, rates were in the range of 9-36%. In the CIS they were generally yet lower, with Russia and Ukraine (at the upper end of the scale) having recipiency rates of 24/25%.

. This can partly

be explained by changes in benefit rules but more by the falling share of short term

unemployed. Further, the impact of an unemployment benefit system depends on the

enforcement of the benefit rules. Evidence that is available, however, relates only to the

strictness of the rules rather than the strictness of actual enforcement. Hungary has had

relatively strict rules throughout while Russia has been less strict. Finally, it should be

noted that there have been changes other than in generosity to benefits systems since the

33

early 1990s. For example, in 1997 the Poles started differentiating over the duration of

benefits by making it conditional on the level of unemployment in a given region.

To explore any possible relationship between unemployment rates and

unemployment benefits, the aim would be to see whether institutions can explain time

evolutions and whether institutions can explain specific country evolutions beyond

common time evolutions and differences in levels. However, given the small number of

observations on time, it is necessary to limit the analysis to some correlations between

unemployment and the NRRs. Charts 3 and 4 now plot the aggregate unemployment rate

of each country against the short and long term NRRs, respectively. Chart 5 also plots the

relationship between unemployment and unemployment benefits using both short and

long term NRRs for both short and long run unemployment. There is no clear, linear

relationship between these indicators and unemployment levels across countries. There is

a slight positive but non-linear correlation between long-term net replacement rates and

both the overall unemployment rate and the proportion of long-term unemployed. The

link for the initial net replacement rates and unemployment is more complex but also

non-linear. There was also no clear linear relationships when controlling for country

fixed effects. In short, unemployment benefits do not appear satisfactorily to explain

unemployment.

However, cross-country variations in unemployment and labour market

institutions are only one way of analysing whether the latter do indeed play a role in

explaining the evolution of unemployment. Using Labour Force Survey (LFS) data

providing information on an individual level for three countries - Poland, Hungary and

the Czech Republic – it is possible to analyse whether the benefit system has had an

34

impact on labour market flows out of unemployment into employment and inactivity over

time. The advantage of this approach is that it is possible to control more fully for

individual circumstances, such as unemployment duration, age and education. To do this,

in addition to the usual control variables, the NRRs were merged into the LFS data. In

particular, the matching took into account the number of children, marital status and

length of unemployment spell. While this necessarily simplifies - net replacement rates

after all are based on average wages and not on actual wages - it introduces further

variation into the data beyond the usual binary information. This way it is possible to link

changes in average levels of benefits over time with individual information.

Multinomial logistic regressions were then estimated for the three countries.

These were done both on a pooled and annual basis with the aim of tracing changes in the

impact of institutions on flows over time39

39 Full results are reported in Commander and Heitmueller (2008)

. For flows from unemployment to

employment in all three countries, the sign on the net replacement rate varied but was

very rarely significant. When controlling for individual attributes, as well as region and

time, there was no evidence that NRRs had any notable impact on flows into employment

from unemployment. For registrations, there was no robust, common effect across

countries. Estimates were also sensitive to whether pooled or time series data were being

used. For Hungary, where there were more measures of benefits, recipients of disability

benefits in particular had a lower chance of moving back into employment from

unemployment. Being on disability benefits in Hungary reduced a person’s re-

employment probability by around 6 percentage points. Finally, when using a dummy

variable for receipt of unemployment benefit (UI) or social assistance (UA) for the three

35

countries, there was no clear effect or difference with almost all coefficients being

insignificant.

Interestingly, the picture looks rather different for flows from unemployment to

inactivity. Net replacement rates and being in receipt of unemployment benefits, as well

as being registered as unemployed, clearly reduce the likelihood of individuals becoming

inactive in the pooled model in both the Czech Republic and Poland. Second, pension,

disability and child benefits significantly increase the chances of becoming inactive in

Hungary. By contrast, the NRR variable – whether entered individually or jointly – is, as

in the other two countries, mostly negatively signed and significant. When dummy

variables for the receipt of benefits or social assistance are included both enter negatively.

In short, a careful look at a number of transition countries which have adopted

systems of unemployment insurance provides little evidence that benefits can explain

differences in aggregate unemployment rates across countries. Matching NRRs to micro-

data, it appears that benefits can affect outflow rates from unemployment, albeit mainly

with respect to flows from unemployment to inactivity. The evidence suggests that these

effects have stayed constant, but small, over time despite the fact that benefit generosity

has declined. Given the change in aggregate unemployment rates and an increased share

of long-term unemployed with limited benefits entitlements, this suggests that the overall

link between institutions and unemployment rates has been weak. The next section now

turns to the design issues connected with the possible introduction of policies for

reducing employed-based risk.

36

7. Design issues for emerging markets

Previous sections have suggested that emerging markets face increased risk from

continuing integration in to the world economy. This risk appears to be associated with a

greater appetite among their citizens for government to provide measures of risk

abatement in the labour market. Yet, even if these preferences are to be met, there is no

clear template in design that emerging markets can easily. Section 3 has shown, for

example, that there are wide differences in the design of unemployment insurance

systems in the OECD as well as very different degrees of employment protection.

Further, the mix of employment protection and unemployment benefits provision varies

significantly across countries.

The available evidence also indicates that the dominant form of public

intervention in the labour market in emerging markets remains employment protection

with, in particular, use of the public sector as a vehicle for employment creation and job

preservation. In part because of these preferences, the size of the formal sector has itself

often remained relatively small. Employment protection has also tended to run alongside

high costs for separation, hence high severance charges. These can impose severe

rigidities on firms and result in both the levels of employment and the composition of

employment being skewed in ways that are adverse from the standpoint of productivity.

Further, while the direct beneficiaries of intervention have generally been a fairly small

share of the labour force, the consequences of the type of intervention have been

encompassing and have helped structure the organisation of the wider labour market. In

this sense, policies aimed at lowering employment based income and consumption risk

require sharp departures from past practice. Blanchard (2004) summarises the challenge

37

as moving away from high severance payments and employment protection more

generally to a system of publicly provided unemployment insurance and lower severance

costs. Others have argued that unemployment insurance in the OECD manner is either

too costly fiscally and/or in terms of incentives and that a greater emphasis on self-

insurance would instead be desirable40

When considering what experience with policies – mainly in the OECD - of

employment-based risk mitigation suggest for future design, several aspects stand out.

There is a vast body of evidence that shows that social protection is costly. Including

unemployment benefits, social protection expenditure accounts for over 25% of GDP in

Western Europe. A body of research has also found that unemployment benefits can

affect not only the rate of unemployment but also that for non-participation. In particular,

there is compelling evidence that the presence of benefits affects the duration of

unemployment, not least as indicated by the jump in job matching that occurs when

eligibility for receipt of benefits expires

. However, in most instances to date, such changes

have proven to be wishful thinking, not least because of the entrenched power of

incumbents and vested interests.

41

40 For example, Vodopivec (2009)

. The presence of unemployment benefits can

affect workers’ incentives to search for work and the scale of these effects can be very

significant. In emerging markets, the question of institutional capacity is also critical.

Most emerging markets lack robust, transparent and effective institutions. This is likely to

be a major constraint in a domain where integrity, monitoring and oversight – not least to

limit moral hazard among recipients of transfers – is essential. In China, for example, the

main challenge facing recently introduced labour laws remains enforcement and

41 See, inter alia, the review in Atkinson and Micklewright (1991)

38

implementation and this is a characteristic feature of most emerging markets42. Further,

the size of the informal economies in most emerging markets is generally very significant

– in India, for example, the relative share of the formal sector in industrial employment is

under 10% 43- and this raises further questions regarding coverage and feasibility.

Moreover, the presence of large informal sectors often implies that unemployment is

harder to isolate as a discrete state, as the response to job loss is for workers commonly to

shift into lower paid informal activity44. Finally, given that the formal sector in most

emerging markets is in effect the privileged sector in terms of compensation, contractual

terms and working conditions, workers losing jobs in the formal sector tend not to be

those with most income risk or exposure to poverty45

. In short, when considering

appropriate design in the emerging market context, there is no unique set of design

principles and practices that can readily be forced into a template.

7.1 Merging savings and insurance?

For some time, mainly with a view to mitigating the moral hazards involved in traditional

systems of unemployment insurance, economists have proposed merging elements of

saving or self insurance with a funded insurance component. For the USA, for example,

Feldstein and others (1998) relied on simulations to argue that UISAs, or unemployment

insurance savings accounts, could significantly lower the cost to taxpayers of providing

unemployment benefits. Others have argued that savings for unemployment and

42 See OECD (2010), p171-2 43 OECD (2007), p121 44 However, this response is of course endogenous in the sense that absent any organised system of fall-backs, workers will have to seek other work or rely on family members. 45 A point made by Vodopivec (2009).

39

retirement need to be merged to be effective in making workers internalise the costs of

unemployment benefits as well as allowing for better diffusion of risk.

In recent years, these proposals have been advanced as a possible solution for

some emerging markets and, indeed, there is now some experience – notably in South

America – of just such applications. It has been argued that their main benefit is to limit

the disincentives for seeking work for those in receipt of transfers. Some element of self

insurance could be expected to improve incentives for unemployed workers to search for

work. Indeed, a recent study of a UISA programme in Chile has found, for example, that

recipients relying primarily on UISAs behave differently than those relying on transfers

with respect to the timing of their exit from unemployment and that this difference is

consistent with improved work incentives46

UISAs have been described in detail elsewhere

.

47

46 Van Ours et al (2010)

. The main features that should

be noted are that employers and in some cases workers deposit a specific share of a

worker’s earnings in an individual savings account. In Latin America – where these

accounts have been most widely used – the deductions that are made have ranged

between 3% to over 9%. In case of job loss, workers can draw on these accounts. The

main difference in the design of UISAs concerns the extent to which redistribution or

borrowing is allowed. Savings accounts that allow borrowing imply that workers are not

necessarily constrained by the volume of their savings once experiencing a bout of

unemployment. In most instances, withdrawals are not simply bounded by the amount in

an account, so that borrowing or recourse to an additional fund can be made for a limited

period and/or amount.

47 See Robalino et at (2009) for a good overview, also Vodopivec (2006)

40

A widely cited example is that of Chile. There, in a system introduced since 2002,

employees and employers have contribution rates of 0.6 and 1.6% respectively. An

unemployment episode can lead to a worker withdrawing from the savings account to a

maximum of five monthly payments with a replacement rate of between 30-50%. If

savings are not adequate to cover this eventuality, the worker can draw on a Solidarity

Fund up to this same limit, but, in this instance, only if job loss was involuntary. Further,

access to Solidarity Fund resources has been restricted to workers with open ended, rather

than fixed term, contracts, although this has now been partly relaxed. The Solidarity Fund

is in turn financed by employer contributions of 0.8% of all wages of their employees

with an additional government annual lump sum contribution. The total contribution rate,

excluding the government element, consequently amounts to 3%.

The Chilean and other examples are cases in which private contributions are

supplemented by public insurance. In the case of a Solidarity Fund, the public insurance

component has, however, an explicit redistributive aspect, although actual

implementation of the system is in the hands of a private firm selected competitively.

Some schemes – as in the case of the system being discussed for adoption in Jordan –

involve borrowing with a key issue being to do with the balance at retirement. In many

emerging markets, average life expectancy remains far closer to retirement age than in

the OECD and this could reasonably be expected to influence behaviour by accelerating

the incentive to borrow towards the end of their working lives. To limit possible abuse,

those with outstanding borrowings are mostly expected to repay any excess borrowing

but there are obvious concerns about the credibility of this provision. In some instances,

scope for forgiveness or partial subsidisation can be introduced.

41

7.2 Limits to self insurance

While UISAs can be motivated on both macroeconomic – boosting savings rates – and

incentive grounds, some fairly fundamental problems of design remain in the context of

most emerging markets. Most generally, such schemes involve high levels of institutional

capacity and integrity, as well as a financial system capable of managing and investing

the array of individual accounts. But there is also the objection that workers – or

particular groups of workers – may be unable to generate sufficient savings to draw down

over an unemployment spell. This is likely to be particularly true for young workers who

commonly face particularly high hazards of unemployment, let alone for low wage

earners in the informal sector. Unlike pensions which get drawn down at a defined time

in life, episodes of unemployment tend to be far less predictable and hence timing

becomes a potentially complicating issue. Moreover, in economies where wage levels are

relatively low, workers may be unwilling to save for events that are hard to predict and to

which they may attach very widely differing probabilities and, hence, priorities. Indeed,

the savings shortfall criticism may hold in aggregate in many emerging markets, as

savings ability over extended periods of time may also be inadequate. That need not, in

principle, imply that the cost of any supplementary public funding would have to be

greater than under an alternative regime, but it does signal that UISAs in emerging

markets are likely to carry a non-trivial, fiscal cost.

Most generally, in terms of design it appears that systems that involve a mix of self

and public insurance might be able to address effectively some of the financing and

equity issues, if not those of institutional capacity. Robalino et al (2009) rightly point out

that conventional UI systems are not necessarily progressive and argue that UISAs can be

42

designed so as to redistribute to poorer workers, such as through limiting benefits and/or

ensuring a minimum level of benefits. However, aside from any difficulty in

implementing this, moving to any system where redistribution (using public funds) can

occur will raise many of the same problems as a publicly funded insurance system, such

as with respect to incentives for job search. Such schemes also do not address the issue of

layoffs and the challenge of trying to make firms internalise the cost of the public

resources used to complement any self-insurance element48

A further – and familiar - criticism of UISAs is that they unsuitable for countries with

large informal sectors. Given that the bulk of the workforce in many emerging markets is

actually in the informal sector, moving to any form of self insurance with or without a

public element would require a sea-change in behaviour among both employers and

employees and could be expected to run into serious enforcement-cum-incentive

problems given the institutional, taxation and other status of informal sector firms and

individuals. Although, in principle, adopting an explicit redistributive objective through

use of public resources could allow extension of coverage to informal workers, it is not

clear whether (a) such an extension of coverage would match to demand, (b) would be

administratively and institutionally feasible and finally, (c) would be free from the usual

design problems facing more conventional programmes. Robalino et al (2009) argue that

it would require a relatively small redistributive component that was well targeted for

incentive and fiscal costs to be contained. But given the scale of the informal or

uncovered sector as well as institutional and information constraints, this argument may

be heroic.

.

48 Blanchard (2004)

43

7.3 Facilitating reform

Before moving to actual design principles, it should also be emphasised there are likely to

be some important political economy considerations. Earlier sections have given details

on the pervasive nature – and corrosive consequences – of the employment protection

that exists in most emerging markets. However, aside from lowering employment risk for

insiders but also reducing job creation and prolonging unemployment durations for

others, employment protection policies have built up substantial groups of interested

parties with an aversion to reform. As such, a major barrier to reform concerns the

bargaining power of those already protected – incumbents. In addition, many of those

protected are employed in public sector firms, government or in large private companies.

Such firms are often those that are highly unionised and where the voice for protection

remains powerful49

49 Union presence and bargaining coverage has also affected the extent and persistence of employment protection in OECD countries. Low union density need not imply low bargaining power but may depend on the structure of bargaining and coverage. The example often cited is that of France where union membership has been low and declining.

. This makes their workers often potentially formidable opponents. Set

against that is, however, the far larger constituency of workers outside the protection

sector, including those queuing to secure jobs in the protected sector. Indeed, the median

voter looks far closer to someone in the unorganised part of the economy in most

emerging markets. However, they tend to form no coherent constituency – not least for

being formed of disparate parts – and hence tend to lack effective bargaining power. In

addition, many governments draw their political support from the protected sector and in

effect pursue their support through providing superior benefits and other employment

based privileges.

44

A common way around these problems has been to try and use severance

payments. While design has varied, most generally such payments have had a lump sum

component and when appropriately discounted have been pitched lower than the NPV of

remaining in protected employment. In some instances, there have been experiments -

similar to those tried in Western Europe - using different types of contracts for new hires

that also involve grandfathering incumbents. However, while severance can have a place

in compensating for job loss – particularly for those who have been in a job for a

protracted period of time - severance is inherently unsuitable as a more permanent way of

dealing with employment and income risk.

The relevant question is what sort of compensation mechanism can assist in

promoting restructuring and a reduction in employment protection in formal sector firms?

To the extent that severance can deal only with a proportion – and possibly a small

proportion – of possible job losers, the introduction of more general employment loss

compensation might be able to advance a more general reform of the labour market. By

providing some form of fallbacks for those with an enhanced employment risk, the

objective would be to enable a wider pursuit of reforms that involve a diminution in

employment protection. To that end, the aim would be to substitute over time

unemployment insurance for employment protection. Indeed, this quid pro quo is

essential if grafting on some element of unemployment insurance to existing systems of

(excessive) employment protection is to be avoided. Firms would in principle gain from

the increase in discretion regarding hiring and firing decisions. At the same time,

reductions in payroll tax rates for employers could be made to reflect a rebalancing of

risk sharing and to increase the attraction of the policy shift.

45

7.4 Elements in design

The above discussion has highlighted several features and constraints. For a start,

classical systems of unemployment insurance are demanding institutionally and generally

are plagued by major incentive problems. They are also costly. Financing through

payroll taxes – rates of which are commonly quite high - also imposes costs, notably with

regard to job creation and in some instances can affect the type of jobs that are created.

Attempts to address incentive issues through self insurance have made some headway but

have serious drawbacks. Not only are there questions of coverage, adequacy and

implementation but the more realistic variants of self insurance that involve elements of

redistribution face many of the same incentive issues of more conventional approaches.

Further, if coverage is to be extended to the informal sector not only is this likely to

require a change in attitudes and modus operandi but also will likely involve major fiscal

costs. Even so, despite these shortcomings, there is clearly scope for meshing together

elements of public and self insurance to address employment-based income risk in

emerging markets: the question is how?

When designing any system of employment risk compensation, three players are

relevant. The first comprises the workers whose employment risk is the primary subject

of the policy. The second comprises the firms who employ those workers and whose

performance and preferences will tend to determine the level and variation of

employment over time. The third comprises government whose ability to fund and

monitor unemployment insurance is, for a variety of reasons, superior to the other

players.

46

In the case of firms, Blanchard (2004) points out that the presence of unemployment

insurance will distort the separation decision. High benefits will affect the productivity

level at which firms are indifferent between employing workers and laying them off. This

distortion can in principle be addressed – in the case of risk neutral firms – by firms being

charged for unemployment benefits paid to workers they lay off, thereby making them

internalise the costs of their decisions. But if the firm is risk averse and faces financial

constraints then this sort of recovery of costs is unlikely to be feasible50

Relying on a larger component of self insurance involves, in effect, workers putting

aside savings to be drawn down in the event of unemployment. This can be funded in a

variety of ways, through a payroll tax contribution, most likely of a compulsory nature, or

through voluntary contribution. Relying, however, on a larger component of self

insurance could also be non-neutral vis-à-vis job creation as it would depend on the level

of deductions from firms. For workers, the level of savings could potentially have an

impact on labour supply. And – as pointed out earlier – if there is an additional element

. In an emerging

market context, particularly where reductions in employment protection could be

expected to induce restructuring, contributions by firms on an ex post basis are unlikely

to work. One possibility would be to use both a mix of fixed contributions – i.e., a payroll

tax – and ex post adjustment with the share financed by payroll taxation relative to that of

the lay off tax being varied over time and context. The underlying principle at work is

that firms who lay off workers should pay some, if not all, of the costs of the benefits

paid to workers who have lost their jobs.

50 Blanchard (2004) makes the point, however, that most lay-offs – at least in the OECD - do not occur when firms are distressed, suggesting that financial constraints may be less binding. But this may not hold for emerging markets, particularly in the context of a reduction in employment protection and associated restructuring.

47

of public insurance, then many of the same incentive or moral hazard issues will be

present.

7.5 Sequence

Consider an archetypical emerging market, likely to be a middle income country with an

existing system of social security contributions and old age benefits (normally organised

around a single state run pillar), alongside a large uncovered informal sector. In these

contexts, when introducing unemployment benefits, at the outset it would be necessary to

limit coverage to workers in the organised or formal sectors with, however, the

establishment of a parallel scheme (of which more below) aimed at extending, in a

staggered manner, coverage towards informal sector workers. Over time, the two might

be expected to converge.

With respect to financing, any viable system is likely to have mixed sources of

funding, as in the Chilean example. There clearly needs to be a public insurance

component, not least because of the ability of the state to diversify unemployment risk51

51 See the wider discussion of this issue in Blanchard and Tirole (2004)

.

In the Chilean example, this comprises the Solidarity Fund which is part covered by

public contributions. There is also scope for an explicit introduction of self insurance.

Given that coverage would start only with the existing formal sector – which in many of

these countries is relatively small – the creation of individual employment accounts could

be feasible. Each worker would have an individual employment account held with the

Employment Agency (but possibly implemented through a private sector operator or

bank). Payments into that account would come from up to three sources. Employers

48

would make regular (monthly) contributions set as a share of the worker’s gross wage.

Workers would similarly be required to set aside explicitly a (lower) share of earnings

that would be paid directly into the account. Subject to some eligibility criteria (see

below), workers would be able to draw on accumulated contributions up to a specified

maximum duration. In the event that combined firm and worker contributions were

insufficient to cover benefits, a supplemental contribution would be made by

government. Over time, an additional component – a balancing item – could be

introduced for individual employers so that at the end of each time period (e.g., every six

months) in the case when the cost of layoffs (as measured by the sum of unemployment

claims) exceeded contributions, firms would have to make supplemental payments - an ex

post adjustment. This would evidently help to reduce any need for public resources. At

retirement age, any positive balance in the account would be transferred to the worker. To

avoid large drawdown close to retirement, a declining age-related scale of permitted

borrowing on any individual’s account would have to be introduced.

The format suggested above would in effect switch payroll taxation to a specific

employment account with a defined set of contingent benefits. The administrative

requirements would not be trivial for this type of set up but neither would they be that

exacting, particularly given that most of these countries already run contributory old age

programmes. Indeed, proposals to launch individual social accounts have been floated in

the context of some OECD countries52

52 For example, in Laroque (2008)

, while a country like Egypt has already started

introducing individual ‘smart’ cards that will contain, inter alia, citizens’ access rights to

49

services and transfers. Technological changes certainly make implementation more

feasible.

Turning to the design of the unemployment programme, two organising principles

would be needed; simplicity in execution and minimisation of work disincentives. It

would be appropriate to think of a flat rate - but experience-related - payment contingent

on involuntary separation. Given the likely inability of any agency in these emerging

markets to monitor effectively whether workers take jobs and leave unemployment, the

receipt of benefits would have to be more tightly bounded, both in terms of generosity

and duration.

In terms of generosity, a small set of simple skills or earnings brackets (say, high,

medium, low) would be pre-determined with possible claimants being grouped into one

or other bracket, adjusted by years in a job (tenure), at time of application. Replacement

rates would have to set at the lower end of the spectrum, particularly for higher ex ante

earnings brackets. For the lowest earnings bracket, the effective replacement rate should

be set higher. Average feasible replacement rates would have to be determined for each

setting and subject to extensive prior scenario testing but might be expected not to exceed

0.25.

Eligibility criteria should be determined conditioned on the length of time in a job and

work histories. At the outset, a minimum period of one or two years in work could be

considered, possibly combined with a minimum level of contributions in a worker’s

account. A limit on the number of times a claim could be made in a given period would

need to be defined (e.g., one unemployment spell every three years).

50

Duration should again be simple with a clear maximum period for receipt. Given that

any such programme would be mainly about compensating for transitory income loss, it

would be sensible to limit to between 3-6 months. This will not address long term

unemployment but that should not be seen as its task.

In the event of involuntary job loss, payments of unemployment benefits would be

organised equivalently to earnings with weekly frequencies rather than as lump sum

payments. Payments would need to be organised around an institution with a national

branch network – such as Post Offices – but it is possible that other technologies, such as

mobile phones, could also be used to make payments (drawing on experience in some

developing countries with money transfers).

7.6 Bringing in the informal sector

Given that informal sector work is often highly variable in hours, effort and earnings and

is difficult to monitor, isolating a unique labour market state - unemployment – is

unlikely to prove feasible. Conditioning a benefit on such a state would therefore be

equivalently infeasible. As such an important objective of policy should be to help induce

informal firms – particularly the larger ones – to shift into the formal sector. Exerting

influence at this margin may motivate more specific interventions but, in general,

protecting workers against employment and income volatility in the informal sector is not

best served by a system of unemployment insurance.

As regards drawing firms into the formal economy, much, of course, depends on

reducing barriers that range from tax levels and predictability to other regulatory costs

and, hence, to improving the benefits to being in the formal sector. However, by raising

51

the value of being formal – at least to workers – in providing some share of benefits

available to formal sector workers, some further incentive for forms to cross-over could

be introduced. For the firm, it depends on whether they attach value to having access to

some unemployment insurance for their workers or merely view it as an additional cost,

given the effectively de-regulated labour market in which they operate as informal sector

firms. In making this choice more attractive to firms, some temporary and selective

financial incentives could be applied. With respect to unemployment contributions, this

might, for example, take the form of a contributions holiday or partial relief for firms,

alongside public supplements to worker contributions for a transitional period of up to

several years. In principle, such public spending could be recouped over time through the

additional fiscal revenues that formalisation could be expected to deliver.

While the sort of incentives described above might affect choices at the margin

between staying informal or moving into the formal economy, the reality is likely to be

that many small informal firms will be unreceptive. Indeed, providing income support –

particularly to the most disadvantaged income groups - might be better achieved through

discrete transfer programmes not necessarily tightly linked to employment status. Even

so, there are obvious benefits to helping individual workers build, where feasible, some

level of precautionary savings. This suggests that the route of augmented and flexible self

insurance might be appropriate. Individuals could be encouraged to set up individual

employment accounts using a wide branch-based network, such as the Post Office. To

encourage participation, a matching or multiplicative contribution from public resources

could be offered with initially that contribution being a multiple of worker or even group

contributions. To avoid firms simply free-riding, the public top-up could, over time, be

52

made conditional on the employer also contributing. Such a scheme would also have to

have very clearly defined savings caps, given the likelihood of relatively low returns to

savings. The scheme would not permit borrowing. As regards access to accumulated

savings, workers could draw down savings in non-lump sum form to cover self-declared

income shortfalls but only once a minimum threshold of savings had been attained and

for a limited duration. Workers would also be permitted to withdraw accumulated savings

on reaching a specified age. Part of the reason for designing it this way would be to

mimic the unemployment benefits framework in the formal sector with a view to raising

demand for firms to shift into the formal economy.

8. Conclusion

Fluctuations in employment and income for workers are a feature of all economies. The

extent to which such fluctuations are addressed through insurance and other mechanisms

varies very widely across countries. These differences are not strictly explained by

income levels alone. Most obviously, the USA operates a very different system of

unemployment insurance than that present in Europe. What is common, however, is that

as countries become richer, both income levels and the structure of risk associated with

those income levels tend to interact in ways that have historically, as well as

contemporaneously, resulted in greater levels of social protection and, in particular, a

larger role for public insurance and action.

This paper has been concerned with understanding what the impact of this greater

level of social protection – particularly in the domains of employment protection and

unemployment insurance – has been on the rich adopters. It has considered the evidence

53

regarding the impact of labour market institutions on equilibrium unemployment and

hence what sorts of lessons might be derived from cumulative experience in these

countries. Using evidence particularly from Europe, it has been shown that the design of

interventions can have major, long lasting and frequently adverse consequences. Yet, in

the midst of large country heterogeneity, certain design principles can be identified that

are of relevance to the growing number of emerging markets whose growth and changes

in citizen aspirations mean that that they must begin to address issues of employment

volatility. Interestingly, using information from a number of so-called transition

economies where, for historical reasons, social protection has been more advanced, no

clear impact of unemployment insurance on unemployment and employment can be

found. There may, however, be some evidence that employment protection exerts a

negative effect on the employment rate.

In this light, the final part of the paper has tried to develop some principles that

can help guide labour market interventions and institutions for addressing unemployment

in emerging markets in ways that avoid many of the pitfalls experienced in the advanced

market economies and are also feasible from both a financing and institutional

perspective. Given large informal sectors and often weak institutions, it is suggested that

a mix of public insurance and self insurance should be used. Explicit employment

programmes and funding should be established. A public insurance element will be

necessary – relying only on employer and employee contributions has some major

drawbacks. This sort of programme can also help economies and policy make the

transition away from regimes of high employment protection and incumbent rent seeking.

For informal sector workers, conditioning transfers on unemployment is not feasible, but

54

nevertheless measures can be taken that can encourage some informal firms to cross into

the formal sector, while also promoting the growth of precautionary savings among

informal sector workers. Finally, the next stage of the work will need to involve detailed

country-specific design and simulations aimed at testing the impact of such innovations

on worker incentives and behaviour, as well as pinning down the detailed fiscal and

financial implications. Institutional capacity and requirement will also need to be

examined closely.

55

References

Addison, J. T., (1996) `Job security provision and unemployment: revised estimates’,

Industrial Relations 35: 585-603.

Atkinson, A.B., (2005), ‘Top incomes in the UK over the 20th century’, Journal of Royal

Statistical Society, 168, part2, pp325-343

Atkinson, A.B., and J. Micklewright, (1991), ‘Unemployment compensation and laboir

market transitions: A critical review’, Journal of Economic Literature, 29, 4,

pp1674-1727

Baker, D, A. Glyn, D. Howell and J. Schmitt, (2005), ‘Labour Market Institutions and

Unemployment’, in ‘Fighting Unemployment: Limits of Free Market Orthodoxy’,

(edited by D. Howell), Oxford University Press

Bartelsman, Eric, Gautier, P and J. de Wind, (2010), ‘Employment Protection,

Technology Choice and Worker Allocation’, IZA Discussion Paper 4895, April

Bassanini, A, L. Nunziata and D. Venn, (2009), ‘Job protection legislation and

prodictivity growth in OECD countries’, Economic Policy

Belot, M. and J.C. van Ours, (2000). Does the recent success of some OECD countries in

lowering their unemployment rates lie in the clear design of their labour market

reforms? IZA Discussion Paper 147

Belot, M and J.C. van Ours, (2001), ‘Unemployment and Labour Market Institutions: An

Empirical Analysis’, Journal of Japanese and International Economics, 15, pp1-

15

Bennmarker, H, E. Mellander and B. Ockert, (2009), ’Do regional payroll tax reductions

boost employment’, Labour Economics, 16, 5, October

Bentolila, S, P. Cahuc, J. Dolado and T. Le Barbachon, (2009), ‘Two-Tier Labour

Markets in a Deep Recession: France vs Spain’, Paper prepared for FEDEA 2009

Annual Conference, Madrid, October

Bentolila, Samuel, Juan Dolado and Juan Jimeno, (2008), ‘Two-tier employment

protection reforms: The Spanish experience’, CESifo DICE Report, 2/2008

Bentolila, S. and G. Bertola, (1990) `Firing costs and labour demand: how bad is euro

sclerosis?’ Review of Economic Studies 57: 381-402.

56

Besley, Tim and Robin Burgess, (2004), ‘Can labour regulation hinder economic

performance? Evidence from India’, Quarterly Journal of Economics, 119, 1

pp91-134

Blanchard, O, (2004), ‘Designing labour market institutions’ , MIT, mimeo

Blanchard, O and J. Tirole, (2004), ‘The optimal design of labour market institutions’,

MIT, mimeo

Blanchard, O, (2004), ‘Reforming labour market institutions: unemployment insurance

and employment protection’, Paper prepared for Barcelona Conference on ‘Post-

Washington Consensus’, September

Blanchard, O and A. Landier, (2002), ‘The perverse effects of partial labour market

reform: fixed duration contracts in France’, Economic Journal, June, pp214-244

Blanchard, O, (1996), ‘The Economics of Post-Communist Transition’, Clarendon

Lectures, Oxford University Press

Blanchard, O. and J. Wolfers, (2000) `The role of shocks and institutions in the rise of

European unemployment: the aggregate evidence’, Economic Journal (conference

papers). 110:C1-33

Boeri, Tito and Pietro Garibaldi, (2007), ‘Two tier reforms of employment protection

legislation: A Honeymoon Effect’, Economic Journal, F357-385

Boeri, Tito, I. Conde-Rioz and V. Galasso, (2003), ‘Protecting against labour market risk:

employment protection or unemployment benefits’, IZA Discussion Paper

Boeri, Tito, A. Brugiavini and L.Calmfors, (2001), ‘The Role of Unions in the Twenty-

First Century’, Oxford University Press

Boeri, Tito and Michael Burda, (1996), ` Active labor market policies, job matching and

the Czech miracle’, European Economic Review, 40: 805-817

Boeri, Tito and Kathy Terrell, (2002), ‘Institutional determinants of labour reallocation in

transition’, Journal of Economic Perspectives, 16

Boeri, Tito (2000), ‘Structural Change, Welfare Systems and Labour Allocation’, Oxford

University Press

Botero, J, S. Djankov, R. La Porta, F. Silanes, (2004), ‘The Regulation of Labour’,

Quarterly Journal of Economics, 119, 4, 1339-1382

57

Brewer, M, (2009), ‘How do income support systems in the UK affect labour force

participation?’, IFAU working paper 27

Calmfors, L. and J. Driffill, (1988). ‘Bargaining structure, corporatism and

macroeconomic performance’, Economic Policy: 14-61.

Comamnder, S and K. Tinn, (2008), ‘Evaluating Doing Business’, London Business

School and LSE, mimeo

Commander, S and A. Heitmueller, (2008), ‘Does unemployment insurance help explain

unemployment in transition countries’ London Business School and EBRD,

mimeo

Daveri, F and G. Tabellini, (2000), ‘Unemployment, Growth and Taxation in Industrial

Countries’, Economic Policy, April, pp49-90

Feldstein, Martin and D. Altman, (1998), ‘Unemployment insurance savings accounts’,

NBER Working Paper 6860

Forslund, A and P. Fredriksson, (2009), ‘Income support systems, labour supply

incentives and employment – some cross country evidence’, IFAU Working

Paper, December

Freeman, Richard, (1998), ‘War of the models: which labour market institutions for the

21st century?, Labour Economics, 5, pp1-24

Gimpelson, V and R. Kapelushnikov, (2009), ‘Employment protection legislation in

Russia: Regional enforcement and labour market outcomes’, IZA Discussion

Paper 4484, October

Gruber, J, (1997), ‘The incidence of payroll taxation: Evidence from Chile’, Journal of

Labour Economics, 15,3, 83

Ham John C., J. Svejnar and K. Terrell, (1999), `Women’s unemployment during

transition: Evidence from Czech and Slovak micro-data’, Economics of

Transition, 7: 47-78

Ham John C., J. Svejnar and K. Terrell, (1998), ` Unemployment and the Social Safety

Net During Transitions to a Market Economy: Evidence from the Czech and

Slovak Republics’, American Economic Review, 88: 1117-1142

International Labour Office, (2010), ‘Accelerating a job-rich recovery in G20 countries’,

Report to G 20 Labour and Employment Ministers, Washington DC, 20-21 April

58

Kugler A and M, (2009), The labour market effects of payroll taxes in a middle income

country: evidence from Colombia’, Economic Development and Cultural Change,

57, January, pp335-358

Laroque, Guy, (2008), ‘Comments on Optimal Household Labour Income Tax and

Transfer Programmes’, Institute for Fiscal Studies, April, mimeo

Layard, R, S. Nickell and R. Jackman, (1991), ‘Unemployment: macroeconomic

performance and the labour market’, Oxford University Press

Lazear, E.P. (1990). ‘Job security provisions and employment’, Quarterly Journal of

Economics 105: 699-726.

Lehmann, Hartmut and Alexander Muravyev, (2009), ‘How important are labour market

institutions for labour market performance in transition countries’, IZA

Discussion Paper 4673, December

Leonardi, Marco and Julian Messina, (2009), ‘The Effect of Employment Protection

Legislation and Financial Market Imperfections on Investment: Evidence from a

Firm Level Panel of EU Countries’, IZA Discussion Paper 4158, April

Ljungqvist, L and T. Sargent, (1998), ‘The European Unemployment Dilemma’, Journal

of Political Economy, 106, 3, pp514-550

Micklewright, J. and G. Nagy (2005). ‘Job Search Monitoring and Unemployment

Duration in Hungary: Evidence from a Randomised Control Trial’, Corvinus

University of Budapest, mimeo

Nickell, S. and R. Layard (1999). Labour market institutions and economic performance.

In (O. Achenfelter and D. Card, eds) Handbook of Labor Economics 3,

Amsterdam, North-Holland.

Nickell,S, L. Nunziata, W. Ochel and G.Quintini (2002) ‘The Beveridge Curve,

Unemployment and Wages in the OECD from the 1960s to the 1990s’, in P.

Aghion et al , Knowledge, Information and Expectations in Modern

Macroeconomics: Essays in Honour of E. Phelps, Princeton

Nickell, S, (2002), ‘A picture of European unemployment: Success and Failure’, mimeo,

LSE, November

Nickell, S., L. Nunziata, and W. Ochel (2005). Unemployment in the OECD since the

1960s. What Do We Know? Economic Journal, 115: 1-27.

59

Nicoletti, G and S. Scarpetta, (2001), ‘Interactions between product and labour market

regulations: Do they affect employment?’ OECD, mimeo

OECD, (2010), ‘OECD Economic Survey: China’, Paris

OECD, (2007), ‘OECD Economic Survey: India’ Paris

OECD (2008). Taxing Wages. Paris

OECD (1994). The OECD Job Study, Evidence and Explanations, Vols I and II, Paris

Orszag, J.M., and D. Snower, (2002), ‘From unemployment benefits to unemployment

accounts’, IZA Discussion Paper 532

Pissarides, C. (2000). Equilibrium Unemployment Theory, 2nd edition, Cambridge, MA:

MIT Press

Puhani, P. A., (2000), ‘Poland on the Dole: The Effect of Reducing the Unemployment

Benefit Entitlement Period during Transition’, Journal of Population Economics,

13, 35-44

Robalino, D, M. Vodopivec and A. Bodor, (2009), ‘Savings for unemployment in good

and bad times: options for developing countries’, IZA Discussion Paper 4516,

October

Rodrik, Dani, (1998), ‘Why do more open economies have bigger governments?’,

Journal of Political Economy, 106, 51, pp997-1032

Saint-Paul, Gilles, (2000), ‘The Political Economy of Labour Market Reforms’, MIT

Press

Salehi-Isfahani, D, (2010), ‘Human Development in the Middle East and North Africa’

Virginia Tech and Brookings Institution, Paper prepared for UNDP HDR, 2010

Salvatori, A, (2010), ‘Labour contract regulations and workers’ wellbeing: international

longitudinal evidence’, IZA Discussion Paper 4685, January

Scarpetta, S, (1996), ‘Assessing the role of labour market policies and institutional

settings on unemployment’, OECD Economic Studies, 26, pp43-98

Snower, D and G. de la Dehesa (editors), (1997), ‘Unemployment Policy: Government

Options for the Labour Market’, Cambridge University Press

Van Ours, J and M. Vodopivec, (2010), ‘Incentive Effects of Unemployment Insurance

savings Accounts: Evidence from Chile’, IZA Discussion Paper 4681

60

Venn, Danielle, (2009), ‘Legislation, collective bargaining and enforcement: Updating

the OECD employment protection indicators’, OECD, Paris

Visser, Jeile (2006), ‘Union membership statistics in 24 countries’, Monthly Labour

Review, January, pp38-49

Vodopivec, M, (2004), ‘Income Support Systems for the Unemployed: Issues and

Options, World Bank, mimeo

Vodopivec, M, (2009), ‘Introducing unemployment insurance to developing countries’,

IZA Policy Paper 6, April

Vroman, Wayne and Vera Brsusentsev, (2009), ‘Unemployment compensation in a

worldwide recession’, Urban Institute and University of Delaware, mimeo, June

World Bank, (2008), ‘Tunisia: Labour Demand, Skills Supply and Employment’, Policy

Note, Washington DC

61

Table 1: Opinions regarding state support for the unemployed: evidence for selected

emerging markets from the World Values Survey, 2005/6

Country Democracy is good Democracy is bad n % good Argentina 6.1 6.3 919 0.95 Brazil 7.8 7.9 1446 0.9 Chile 7.7 6.5 895 0.93 China 8.3 7.3 1233 0.94 Egypt 6.7 6.5 3028 0.98 India 8.4 7.4 1527 0.73 Indonesia 4.4 5.4 1894 0.97 Jordan 7.5 6.1 1423 0.76 Morocco 8.1 7.7 1047 0.96 Peru 6.3 6.3 1392 0.89 South Africa 8 6.7 2840 0.9 South Korea 7.1 7.2 1199 0.77 Thailand 6.6 6.4 1528 0.93 Uruguay 7.2 5.9 931 0.91 Comparators France 6.6 6.3 974 0.9 USA 5.7 5.9 1194 0.86 No of respondents 26130 2654 28784 0.91 Source: World Values Survey, 2005/2006 round Notes: Question 151: I am going to describe various types of political systems and ask what you think about each as a way of governing this country. For each one, would you say it is a very good, fairly good, fairly bad or very bad of governing this country. Question 155: Many things may be desirable, but not all of them are essential characteristics of democracy. Please tell me how essential you think ‘state support for the unemployed ‘ is as a characteristic of democracy. Use this scale where 1 means “not at all an essential characteristic of democracy” and 10 means it definitely is “an essential characteristic of democracy”.

62

Table 2: OECD average unemployment rates, 1960-2007, 2008 and 2009

Country 60/69 70-79 80-89 90-99 00-07 2008 2009

Australia 1.8 3.7 7.6 8.8 5.7 4.2 5.6 Austria 1.9 1.6 3.3 3.8 3.5 3.9 5.0 Belgium 2.2 4.4 11.4 11.3 7.4 7.0 7.9 Canada5.0 6.6 9.4 9.6 6.9 6.1 8.3 Denmark 1.5 4.1 8.1 7.7 4.8 3.3 6.0 Finland 1.9 3.6 4.9 11.9 8.6 6.4 8.2 France 1.5 3.5 8.3 10.2 8.6 7.9 9.4 Germany 0.8 2.4 6.1 7.8 9.3 7.3 7.5 Greece 5.3 2.3 6.6 9.6 9.7 7.7 n.a Ireland 5.3 7.2 14.1 12.7 4.3 6.0 11.8 Italy 5.2 6.4 10.2 11.3 8.4 6.8 7.8 Japan 1.3 1.7 2.5 3.0 4.7 4.0 5.1 Netherlands 0.8 3.7 9.9 6.1 3.4 2.8 3.5 New Zealand 0.2 0.5 4.6 7.9 4.5 4.2 6.1 Norway 1.0 1.6 2.8 4.9 3.8 2.5 n.a Portugal 2.5 4.7 7.4 5.6 6.2 7.8 9.6 Spain 1.5 4.3 17.9 19.8 10.6 11.4 18.1 Sweden 1.7 2.1 2.8 7.5 6.2 6.2 8.3 Switzerland 0.01 0.2 0.6 3.0 3.5 3.5 4.4 UK 1.6 3.6 9.8 8.1 5.1 5.6 n.a USA 4.8 6.2 7.3 5.7 5.0 5.8 9.3 Source: OECD

63

Table 3: OECD Overall Employment Protection indicators for 2008

Western Europe Emerging Markets

Austria 1.93 Brazil 2.75 Belgium 2.18 Chile 2.65 Denmark 1.5 China 2.65 Finland 1.96 India 2.77 France 3.05 Indonesia 3.68 Germany 2.12 Korea 1.9 Greece 2.73 Mexico 3.13 Ireland 1.11 South Africa 1.25 Italy 1.89 Turkey 3.72 Netherlands 1.95 Norway 2.69 Portugal 3.15 Spain 2.98 Transition Countries Sweden 1.87 Czech Republic 1.96 Switzerland 1.14 Hungary 1.65 UK 0.75 Poland 1.9 Russia 1.92 North America Estonia 2.1 USA 0.21 Slovakia 1.44 Canada 0.75 Slovenia 2.51 Australasia and Japan Japan 1.43 Australia 1.15 New Zealand 1.4 Source: OECD (2008)

64

Table 4: Unemployment rates in selected transition economies, 1992-2008

Country 1992 1996 2000 2004 2008

Albania 26 12 16.8 14.4 12.7 Bulgaria 15.3 13.5 16.3 12 5.6 Czech Republic 2.6 3.9 8.8 8.3 4.4 Estonia 3.7 9.9 13.6 9.7 5.5 Hungary 9.8 9.9 6.4 6.1 7.8 Latvia 2.3 20.6 14.4 10.4 7.5 Lithuania 3.5 16.4 16.4 11.4 5.8 Poland 13.6 12.3 16.1 19 7.1 Romania 8.2 6.7 7.1 8 5.8 Russia 5.2 9.7 9.8 7.8 6.4 Slovakia 11.4 11.3 18.6 18.1 9.5 Ukraine 0.4 7.6 11.6 8.6 6.4

Source: Lehmann and Muravyev (2010) who use ILO, IMF and Transmonee data. Note that 1992 rates are mostly for registered unemployment, while other years are ILO rates.

Table 5: Net replacement rates for transition economies for initial (I) and long term (LT) unemployment in 2008 (for single individuals and one earner married couples with 2 children at 100% of average wage) Country Single/no children One earner couple+2 children

I LT I LT Bulgaria 50 16 55 43 Czech Republic 53 30 61 57 Estonia 54 18 57 35 Hungary 59 23 70 59 Latvia 83 23 75 28 Lithuania 61 16 63 59 Poland 45 24 46 44 Romania 42 8 44 24 Slovakia 65 19 61 39 Slovenia 64 33 86 78 Comparators France 66 34 71 54 Germany 60 36 72 63 Netherlands 73 61 75 83 Sweden 50 44 60 65 UK 38 38 69 69 USA 55 6 51 38 Source: OECD (2008)

65

0

50

100

150

200

250

0 20 40 60 80 100 120 140

Chart 1. Selected Emerging Economies: change in trade openness from 1980-84 to 2004-2008

Trade openness 2004-08

Trade openness 1980-84

MAL

JORTHL

TUN

EGYKEN

PHLMOR

SAFIDNCHLNIG

URGCHN

PAKCOLTUR

BRA

IND

Source: World Bank's World Development Indicators and authors' estimates

0

10

20

30

40

50

60

70

0 0.5 1 1.5 2 2.5 3 3.5

Chart. Selected OECD countries: OECD Employment protection index and net replacement rate for single person at 100% of average wage

Net replacement rate for single person at 100% of average wage

OECD Employment protection indexUSA

CAN

GBR

SUI

IRL

JAP

DEN NED

AUTBEL

SWE FIN

GER

NOR

FRA

ESP

POR

66

Chart 3: Net replacement rates long term (averages, max and min)

.2.4

.6.8

0.2

.4.6

.2.4

.6

.2.4

.6.8

0.0

5.1

.15

.2

0.5

11.

5

1990 1995 2000 2005 1990 1995 2000 2005 1990 1995 2000 2005

1990 1995 2000 2005 1990 1995 2000 2005 1990 1995 2000 2005

Czech Hungary Poland

Romania Russia Slovak

Mean long-term Max long-termMin long-term

year

Graphs by country

Chart 4: Net replacement rates initial phase (averages, max and min)

.2.4

.6.8

1

0.5

1

.2.4

.6.8

.2.4

.6.8

1

0.5

1

.4.6

.81

1.2

1990 1995 2000 2005 1990 1995 2000 2005 1990 1995 2000 2005

1990 1995 2000 2005 1990 1995 2000 2005 1990 1995 2000 2005

Czech Hungary Poland

Romania Russia Slovak

Mean short-term Max short-termMin short-term

year

Graphs by country

67

Chart 5: Correlations of unemployment and unemployment benefits

0.0

5.1

.15

.2

0 .2 .4 .6 .8Mean NRR long-term

aggregate unemployment ratepredicted unemployment

0.0

5.1

.15

.2

.3 .4 .5 .6 .7 .8Mean NRR short-term

aggregate unemployment ratepredicted unemployment

.1.2

.3.4

.5.6

0 .2 .4 .6 .8Mean NRR long-term

share of long-term upredicted u_long

.4.5

.6.7

.8.9

.3 .4 .5 .6 .7 .8Mean NRR short-term

share of short-term upredicted u_short